Masayoshi Son has a new idea: use robots to build the data centers that run AI, bundle those assets into a single company, and take it public at $100 billion.
SoftBank Group is planning to create and list a standalone AI and robotics company in the US called "Roze," targeting a valuation of about $100 billion and aiming to pursue an initial public offering as early as the second half of 2026. The new entity will focus on building data centers and using robotics to improve the efficiency of AI infrastructure construction. CNBC
KPMG has been hired to prepare the necessary financials and documents for the public offering. SoftBank is planning to host an analyst day at a Texas data center in July to build investor interest ahead of the IPO. Benzinga
What Roze Actually Is
Roze could bundle existing energy, land, and infrastructure assets from SoftBank's portfolio, as well as ABB Robotics, which SoftBank agreed to buy last year. SoftBank is expected to integrate ABB's robotics hardware with AI software capabilities. CNBC
The pitch is straightforward: AI data center demand is growing faster than humans can build the facilities to house it. Roze would deploy autonomous robots to accelerate construction timelines and reduce costs — addressing the physical bottleneck that every major cloud provider cited in their earnings calls this week.
The Skepticism Inside SoftBank
The report noted that the valuation target and timeline could shift, and that the plans are considered ambitious by some SoftBank executives, in part due to uncertainties stemming from the conflict in the Middle East. CNBC
The financial rationale is also visible. SoftBank has committed more than $30 billion to OpenAI. Investors have raised concerns about how SoftBank will fund those investments, particularly as OpenAI remains unprofitable. A public listing for Roze would help SoftBank bring in outside capital to offset some of that pressure. CNBC
The Bigger Picture
Son is betting on a thesis that is genuinely sound even if the timeline is aggressive: the constraint on AI scaling is no longer just compute — it is the physical construction of the buildings that house compute. Whether Roze can execute on that thesis at $100 billion is a separate question from whether the thesis is right. Right now the thesis is right.
